摘要
为解决煤矿机电设备传统维护方法存在效率低、成本高的问题,提出了一种基于深度学习的智能化故障诊断预测方法。该方法集成卷积神经网络(CNN)、长短期记忆网络(LSTM)和门控循环单元(GRU),并引入卷积核注意力机制动态优化特征提取。利用CNN提取振动信号的局部空间特征,LSTM捕捉长期时序依赖,GRU增强短时序列处理能力,结合注意力机制自适应融合多尺度特征,并通过Softmax完成故障分类。将所提模型应用于实测的煤矿机电设备滚动轴承振动信号中,并和其他模型进行对比。结果表明,不同负载下所提出模型的平均故障诊断准确率为99.46%,优于CNN-LSTM、LSTM和DNN,这为煤矿机电设备的智能化运维提供了技术支撑。
In order to solve the problems of low efficiency and high cost of traditional maintenance methods of mechanical and electrical equipment in coal mines,an intelligent fault diagnosis and prediction method based on deep learning was proposed.The method integrates Convolutional Neural Network(CNN),long short-term memory network(LSTM)and gated recurrent unit(GRU),and introduces the convolutional kernel attention mechanism to dynamically optimize feature extraction.CNN is used to extract local spatial features of vibration signals,LSTM captures long-term temporal dependence,GRU enhances short-time sequence processing capability,adaptively fuses multi-scale features with attention mechanism,and completes fault classification through Softmax.The proposed model is applied to the measured vibration signal of the rolling bearing of coal mine electromechanical equipment,and compared with other models.The results show that the average fault diagnosis accuracy of the proposed model is 99.46%under different loads,which is better than CNN-LSTM,LSTM and DNN.This advancement provides technical support for intelligent operation and maintenance of coal mine mechanical and electrical equipment.
作者
王明
刘建庄
WANG Ming;LIU Jianzhuang(Guoneng Shendong Coal Group,Ordos 017200,China;North China University of Science and Technology,Tangshan 063210,China)
出处
《机械与电子》
2025年第8期67-72,80,共7页
Machinery & Electronics
基金
河北省自然科学基金项目(23HBZ085703)。
关键词
煤矿机电设备
深度学习
故障预测
mechanical and electrical equipment in coal mines
deep learning
failure prediction
作者简介
王明(1984-),男,河南平舆人,硕士,工程师,研究方向为信息化、智能化等;刘建庄(1976-),男,河北唐山人,博士,副教授,研究方向为机械控制等。
